Addition Through Subtraction: How a Few Bad Apples Can Destroy Campaign Performance
Conventional wisdom suggests that programmatic advertising benefits from increased exposure to inventory and publishers. Makes sense, right? Supply is a proxy for users, and programmatic is about reaching the right user at the right time with the right message. But it’s also true that not all publishers are the same and campaign performance will vary by campaign, advertiser, creative, seasonality, time of day, etc., across publishers. Supply decisions are a critical element of campaign performance, but one of the things we have learned at MBuy is that the conventional wisdom used to identify high performing publishers frequently eliminates sites that are actually helping to drive performance. To be clear, this isn’t about brand safety – this is about otherwise quality inventory and metrics that are used to differentiate campaign drivers from campaign killers.
View and click-based conversion credit is the most often used metric in differentiating “good” publishers from “bad” in direct response campaigns. But what we’ve found is that this isn’t good enough because ads on sites that get view based conversion credit are frequently never actually seen by the user. In turn, platform buying algorithms start aiming more spend at these sites despite the fact that ad quality is low. Additionally, feeder sites (sites with high viewability that drive future conversions credited to other sites) are ignored and ultimately removed from campaigns which hurts campaign performance overall. Using conversion credit as the primary means of determining site quality can ultimately result in reduced campaign performance.
Because not all impressions are equal, a better way to prioritize sites is to focus on the impression quality that sites can deliver. Impression quality is determined by:
- % viewable impressions (vRATE)
- % viewable conversions (vCONV)
- Cost per viewable conversion
MBuy works closely with companies like AdYapper to provide measurement technologies that enable deeper insights into impression quality across sites. By leveraging these technologies to focus on sites with high impression quality, MBuy achieves significant performance boosts in campaigns.
As an example, MBuy has a travel client that was looking to drive conversions with a CPA under $40. The programmatic campaign launched and in the first two weeks, buys were made across 120,000 sites. MBuy blacklisted 5,200 sites with more than 1,000 impressions, vRATEs under 40%, and 0% vCONV. In other words, sites that were consuming spend but low ad viewability and conversion credits where the user never actually saw the ad. In fact, many of the sites that were eliminated actually had significant view based conversions, but the data showed that they were never actually seen. As a result, impression volume dropped 22%, overall vRATE increased by 26%, clicks increased 43%, CTR increased 83%, and best of all conversions increased by 244% and cost per viewable conversion decreased by 77%.
How is it possible that by removing 4.3% of sites that raw conversions increased so significantly? Think of it as opportunity cost. By eliminating the resource drains on performance, the system was able to better spend budget on higher quality impressions. In fact, of the 5,200 sites that were removed, the majority of performance gains were seen simply by removing the top 100 “bad” sites. In this way, it is truly addition by subtraction. So the next time you think about optimization, keep in mind that a few bad apples can have a significant negative impact on campaigns.